HC3 human / ChatGPT answer classifier
Base model: sentence-transformers/all-MiniLM-L6-v2. Labels: 0 = human, 1 = ChatGPT. Pinned HC3 revision: 4d0ff18143b5a7e1b1e79beb540c04549d1e59d3. Question-disjoint, balanced 80/10/10 train/validation/test split with seed 42. Five training epochs with AdamW, learning rate 2e-5, batch size 32, max length 256.
Baseline test accuracy: 0.844901. Fine-tuned test accuracy: 0.991431. Mean training loss: 0.022282.
This historical benchmark does not establish reliability for current student work, other generators, or other domains. False positives are possible.
- Downloads last month
- 5